The PUSH, QUEUE, and EXECUTE commands are especially useful when used in conjunction with the pause and resume features because they enable you to generate a pause-interrupt command to execute the code you push and return from it via a pushed RESUME statement. In fact, this is precisely how the EXECUTE subroutine is implemented generally.

Caution: Note that the push and resume features work this way only in the context of being inside modules. You cannot resume an interrupted sequence of statements in immediate mode—that is, not inside a module.

For example, suppose that you collect program statements in a matrix called CODE. You push the code to the command input stream along with a RESUME statement and then execute a PAUSE statement. The PAUSE statement interrupts the execution, parses and executes the pushed code, and returns to the original execution via the RESUME statement. Here is the code:

   proc iml;
   start testpush;
      print '*** ENTERING MODULE TESTPUSH ***';
      print '*** I should be 1,2,3: ';
      /* constructed code * /
      code = ' do i = 1 to 3; print i; end;   ';
      /* push code+resume */
      call push (code, 'resume;');
      /* pause interrupt  */
      pause;
      print '*** EXITING MODULE TESTPUSH ***';
   finish;

When the PAUSE statement interrupts the program, the IML procedure then parses and executes the following line:

   do i=1 to 3; print i; end; resume;

The RESUME command then causes the IML procedure to resume the module that issued the PAUSE.

Note: The EXECUTE routine is equivalent to a PUSH command, but it also adds the push of a RESUME command, then issues a pause automatically.

A CALL EXECUTE command should be used only from inside a module because pause and resume features do not support returning to a sequence of statements in immediate mode.